I have just read this post for the first time, probably quite unusual for me. I have been out of town attending some very interesting technical conferences.
I am intrigued at the level of diagnostics that have evolved on this problem and I wonder if all the audience thinks that techs use such technique.
I haven't heard one level of diagnostics used here except for fault codes. We have no comment to what actually is happening. Do we have single cylinder misfires? We have no information on secondary voltages and burn times. All available with a competent scanner and probably a key to the idle answer.
Is closed loop activity initiated soon after cold start? What are the adaptive values at idle, lower and upper partial load?
Parts swapping is definitely part of these diagnosis. But that should be the job of a technician not the customer. Air Mass Meters almost always have to be swapped to prove a defect. The adaptive values are watched before and after and the car is used for the test. Small dynamic defects are compensated for by the management software and even MB says that scan data can not predict AMM problems except through fault code (major problems - not performance problem) Change in adaptation values combined with a hypothesis that explains symptom activity, is the diagnostic concept.
My first guess would have been the coil connectors and the other ends might as well be shot gunned next. If that doesn't do the job, a proper analysis might be the next answer. Start with someone with a scanner that will give the secondary firing voltages and burn tiimes. It gives him a quick window.
Bosch Master, ASE Master, L1
33 years MB technician